Estimation of the duodenal flow of microbial nitrogen in ruminants based on the chemical composition of forages: a literature review
Estimation of the duodenal flow of microbial nitrogen in ruminants based on the chemical composition of forages: a literature review
Jules M.J. GOSSELINK 0 1
Claude PONCET 1
Jean-Pierre DULPHY 1
John W. CONE 0
0 ID TNO Animal Nutrition , PO Box 65, 8200 AB Lelystad , The Netherlands
1 INRA, Centre de Clermont-Ferrand-Theix, Unité de Recherches sur les Herbivores , 63122 Saint-Genès-Champenelle , France
- The objective of this study was to evaluate the estimation of the duodenal flow of microbial nitrogen (N) in ruminants fed forage only, per kilogram of dry matter (DM) intake, which is the yield of microbial protein (YMP). The estimation was based on the chemical composition of forages. A data file of 62 observations was collected from in vivo studies on cattle and sheep fed diets with forage only. A statistical analysis of YMP was conducted with neutral detergent fibre (NDF), crude protein (CP), non structural carbohydrates (NSC), group of forage species (legumes or grasses), method of conservation, physical form of presentation, level of DM intake, animal species, methodology and references as parameters. After a stepwise regression, CP was significant and the most important predictor. NSC or the method of conservation had an extra effect on YMP. On the basis of these three parameters the best fit equations were found and the influence of all parameters on YMP were discussed. Using the data file of this study, the prediction of YMP from the PDI-system was also validated. The statistics of the validation of the PDI prediction were similar to the statistics of the equations from this study. In conclusion, the chemical composition of forages, with or without the method of conservation, is a poor indication for the duodenal flow of microbial N (g·kg-1 DM intake) in ruminants fed diets with forages only.
de matière sèche (MS) ingérée, c’est-à-dire le rendement en azote microbien (RNM). La prévision a
été basée sur la composition chimique des fourrages. Un ensemble de données comprenant 62
observations a été constitué en sélectionnant les études in vivo sur bovins et moutons alimentés avec un seul
fourrage. L’analyse statistique a porté sur la relation entre RNM et différents paramètres : teneur en
glucides pariétaux (NDF), teneur en Matières Azotées Totales (MAT), teneur en glucides
non-pariétaux (GLU), famille botanique (légumineuses ou graminées), méthode de conservation, forme
physique de présentation, quantité de MS ingérée, espèce animale, méthode de mesure et références.
Après la régression « stepwise », l’effet de la teneur en MAT a été significatif et le plus important.
GLU comme la méthode de conservation ont eu un effet supplémentaire sur RNM. À partir de ces
3 paramètres, les équations de prédiction sont proposées. Les paramètres statistiques des équations et
l’influence des différents critères de prévision du RNM sont discutés. À partir de cette base de
données, la prévision du RNM du système PDI était validée aussi. Les paramètres statistiques de la
validation du système PDI étaient similaires aux paramètres statistiques des équations de cette revue
bibliographique. En conclusion, la composition chimique d’un fourrage, avec ou sans la méthode de
conservation, est une pauvre indication pour le flux duodénal d’azote microbien (en g·kg–1 MS
ingérée) chez le ruminant recevant une ration composée uniquement de fourrage.
rumen / azote microbien / légumineuses / graminées / prévision
The objective of this literature review
was to evaluate the protein digestion in
ruminants measured by in vivo experiments.
This evaluation was done as a part of the
revision of the feed protein evaluation system
in France, PDI [
]. The amount of
microbial protein synthesised in the rumen is of
importance in this system and is on average
64% of the flow of protein to the duodenum
in ruminants consuming forage diets. The
quality of microbial protein is quite
constant and high because of their amino acid
]. However microbial protein
flowing out of the rumen can vary,
depending on factors like forage species,
physiological stage, method of conservation and
physical processing of forages .
Microbial protein flow has been
predicted by the daily intake of dry matter
(DM) or organic matter (OM) [
9, 44, 49
more precisely, based on an index of
organic matter fermented in the rumen
(FOM), which is used in the French
] and the Dutch DVE/
]. However the intake of
DM or OM is a rough predictor, FOM is
estimated from OM digested in the total
digestive tract and both predictors comprise
rumen available nitrogen as well as
Microbial growth depends on the
amount and availability of nitrogen and
energy, supplied by the non structural and
structural carbohydrates in feed [
Structural carbohydrates can be
represented by neutral detergent fibre (NDF) and
has supplemental effects on microbial
growth in the rumen . NDF content in
feed DM also affects the rate of
carbohydrate digestion, which is the major factor
controlling the amount of energy available
for microbial growth in the rumen [
A lower NDF content is accompanied by
higher concentrations of non structural
carbohydrates (NSC) and crude protein (CP).
CP favourably improves the efficiency of
microbial growth as long as nitrogen is not
limiting and protein is not used as a source
of energy [
When contributions of these different
chemical components of forage DM (CP,
NDF and NSC) to the synthesis of
microbial protein are known, the estimation of the
duodenal flow of microbial nitrogen (N)
can be made. The importance of NDF in the
estimation of the duodenal flow of
microbial N has been shown by Oldick et al. [
who estimated the daily flow of microbial
N to the duodenum on the base of DM
intake and NDF content. Because DM intake
explains the major part of the daily
duodenal flow of microbial protein [
prediction of this flow will be more refined
when it is estimated per kilogram of DM
The estimation of the duodenal flow of
microbial N in ruminants, fed forages only,
from the chemical composition of forages
and in gram per kg of DM intake is another
approach compared to the calculations of
the flow of microbial N from the PDI– or
]. The objective
of this study was to evaluate this approach
and to validate the calculations from the
PDI-system, using a database from the
literature. Because concentrates or ground
forages have a great effect on the duodenal
flow of microbial protein [
selected in vivo data were from diets
containing chopped or long forages only. The
duodenal microbial flow per kg of DM
intake is called hereafter the yield of
microbial protein (YMP).
2. MATERIALS AND METHODS
2.1. Data file generation
A data file containing 62 observations
was generated from 34 studies published
during the last thirty years [
2, 3, 5–8, 16, 17,
20, 21, 24–26, 28–36, 39, 41–43, 45, 46, 52,
53, 55, 56, 60, 61
]. The 62 observations
contain 27 observations with legumes
(lucerne: 19 and clovers: 8) and 35
observations with grasses (Lolium perenne: 14,
Dactylus glomerata: 4 and other grasses:
The experiments with sheep and cattle
with cannula in the rumen and in the
abomasum or in the proximal duodenum
and with a clear description of the
experimental conditions were selected. All
selected publications contain data of the flow
of microbial N to the duodenum and the
chemical composition of feed DM, at least
CP (g·kg–1 DM) and NDF (g·kg–1 DM). The
determination of NDF was done according
to the different techniques of Van Soest
et al. [
23, 48, 58, 59
] and the determination
of CP was done with the Kjeldahl method.
Non structural carbohydrate (NSC, g·kg–1
DM) was calculated as OM minus CP
minus NDF. As a consequence of this
calculation, NSC also comprise low concentrations
of lipids , which have a small contribution
to the energy delivered to microbial
Other parameters, which might have an
effect on YMP and which were clearly
described in the publications, were also
collected for the estimation of YMP in
addition to the main chemical components
(CP and NDF) in the analyses (Tabs. I and
II). The forages were grouped in legumes
and grasses and were not represented by the
forage species in the analyses because of
the low numbers of data for each species.
Data on the method of conservation (fresh,
hay or artificially dried forage and silage),
physical form of presentation (chopped or
long), the level of dry matter intake (DMI, g
DM·kg–1 body weight) and animal species
(sheep or cattle) were also collected. The
stage of maturity, which is a characteristic
of the forages, could not be used in the
analyses, since it was not given precisely in the
publications. However, the chemical
composition of forages are well related to the
stage of maturity of the forages [
2.2. Description of the data file
The chemical components (CP, NDF
and NSC) well differentiated legumes and
grasses (Tab. II). Although the ranges of
these chemical components in the groups of
legumes and grasses were wide, the values
in the ranges were continuously distributed.
However, the analysis of the difference
between these two groups of forages might be
biased by the parameter animal species,
because experiments on legumes were mainly
done with sheep and experiments on
grasses with cattle (Tab. I).
On the contrary to the duodenal flow of
non ammonia N per kilogram of DM intake
(NAN), the duodenal flow of microbial N,
expressed as YMP and as EMPS (efficiency
of microbial protein synthesis: g duodenal
flow of microbial N per kg OM apparently
digested in the rumen), was significantly
different between legumes and grasses
(Tab. II). The mean values of YMP and
EMPS in the data file were lower for
grasses than for legumes. The variation in
YMP was less large than the variation in
] was used to statistically
analyse the data file and to find the best fit
equation for the estimation of YMP and
NAN from the chemical composition and
the other collected parameters. The
parameter method of conservation (MC)
contained only 2 classes, fresh forages and
others, because YMP was significantly
different (P < 0.05) between fresh forages and
other methods of conservation, but no
significant differences were found between
the other methods of conservation in the
range of NDF content of 400 to 550 g·kg–1
DM (Mean values for YMP (± SE) were:
15.4 (1.27) for fresh forages (n = 8), 12.0
(0.96) for hay and dried forages (n = 11) and
11.9 (0.93) for silage (n = 12)). NAN was
not significantly different for these
methods of conservation.
To account for the variation among
experiments or studies used in the data file,
the parameters methodology and
references were included in the analyses. In the
analysis of YMP, 4 classes of methodology
were composed on the basis of the marker
to measure microbial protein and on the
basis of the method of measurement of the
duodenal flow, with one or two flow markers
and with a different type of duodenal
cannula (Tab. III). In the analysis of NAN,
3 classes of methodology were composed
on the basis of the measurement of the
duodenal flow (Tab. III). The parameter
references (n = 34) represent the 34 studies used
in the data file.
At first the RCHECK procedure of
GenStat was used to check the normal
distribution of the data in the file. The
correlation coefficients between the chemical
components, the other parameters, YMP,
NAN, DM intake per day (DMd) and the
duodenal flow of microbial N per
day (Mday) were calculated with the
a Not significant (P > 0.1); CP: crude protein, NDF: neutral detergent fibre, NSC: non structural carbohydrates,
DMI: dry matter intake, YMP: yield of microbial protein, EMPS: efficiency of microbial protein synthesis,
NAN: non ammonia N.
Candidate equations to estimate YMP
were found by using stepwise regression
and the FIT procedure. To reduce
overparameterisation and multicollinearity
in the model, two selections of predictors
were done before the regression procedure.
At first, the candidate models were
composed from the chemical components and
their quadratic terms, using the RSELECT
procedure. This procedure calculates the
Mallow Cp and selects predictors on the
base of the residual sum of squares and the
number of predictors. Secondly, the other
parameters were added individually to the
candidate models using the FIT procedure
to find out which parameters and
interactions could be significant in each candidate
Yijklmno = β0 + β1Ci + β2Dj + Ek + β3CDl
+ β4 CEm + β5DEn + ε ijklmno
where Yijklmno = YMP or NAN; Ci or Dj =
chemical components, NDF (g·kg–1 DM),
CP (g·kg–1 DM) or NSC (g·kg–1 DM); Ek =
one of the parameters (group of forage
species, method of conservation, physical form
of presentation, animal species,
methodology, DMI or references); CDl, CEm and
DEn = interactions between chemical
components and the added parameter; β0 to 5 =
regression coefficients; ε ijklmno = residual
A stepwise regression analysis of YMP
and NAN was done using the candidate
models with the chemical components, using the
Purine in digesta
Amino acid profile,
For abbreviations, see Table II.
parameters, which were significant in
model 1, and using the parameters, which
had a significant interaction with a
chemical component in model 1.
β0 +β1Ci + β2Dj + Ek + Fl + β3CDm
+ β4 CEn + β5DEo + β6 CFp + β7DFq
+ β8EFr + ε ijklmnopqrs
where Yijklmnopqrs = YMP or NAN; Ci or Dj =
chemical components, NDF (g·kg–1 DM),
CP (g·kg–1 DM) or NSC (g·kg–1 DM); Ek or
Fl = parameters (group of forage species,
method of conservation, physical form of
presentation, animal species, methodology,
DMI or references); CDm, CEn, DEo, CFp,
DFq, EFr = interactions between chemical
components and parameters; β0 to 8 =
regression coefficients; and ε ijklmnopqrs = residual
Overparameterisation was reduced using
only two-way interactions. Multicollinearity
in the final candidate models was evaluated
by calculating the contribution of each
variable to the sum of the squares (regression).
Based on these procedures, candidate
equations to estimate YMP and NAN were
composed. R2 (determination coefficient)
and the probabilities of the equations and
the estimates were calculated.
The difference between the observed
and predicted (estimated) flows was
calculated as the mean square prediction
error (MSPE), according to Bibby and
MSPE = 1/n Σ( O-P)2
O is the observed value and P is the
predicted value and n is the number of
observations. The square root of MSPE expressed
as the percentage of the observed mean is
used as a measure of the prediction error.
MSPE was decomposed into the error in
central tendency (bias), error due to
regression (deviation from regression being one)
and error due to disturbances (unexplained
These statistical parameters were used
to find the best fit equations out of the
candidate equations. A decreased R2 and an
increased prediction error of the predictions
of YMP and NAN could be expected,
because of the high number of variation
factors and the small number of available data.
Therefore, the best fit equations were
also compared according to a method
proposed by Mitchell [
]. The essence of this
method is that 95% of the deviations,
calculated as predicted minus observed values,
are within the envelope of acceptable
precision. The limits of this envelope can be
defined with reference to the purpose of the
model. In this study, SD (standard
deviation) of YMP and NAN in the data file were
used as limits. Also the limits 1.2*SD and
1.5*SD were used, because it is
unreasonable to expect the model to perform as well
as the in vivo data [
a Not significant (P > 0.05); for abbreviations, see Table II.
The duodenal flow of microbial N per
day was correlated with the daily dry matter
intake (Tab. IV). In the statistical analysis
of this flow, the parameters, references or
methodology, were significant (P < 0.05).
These parameters were also significant
(P < 0.001) in the analysis of NAN, which
was correlated with CP (Tab. IV). Because
these parameters were not significant in
models to predict YMP, the results are
focussed on YMP.
YMP was normal distributed and had
the highest correlation coefficients with the
chemical components, CP, NDF and NSC
(Tab. IV). The candidate models for the
estimation of YMP were based on CP or CP2,
with or without NDF, NDF2, NSC or NSC2
(Tab. V). NDF and NSC, which were
correlated, could replace each other. NSC would
be more supplemental to CP in the
prediction of YMP, because the correlation
coefficient between CP and NSC was lower than
between CP and NDF.
In the candidate models with CP2 or CP
plus CP2 the parameter, method of
conservation, tended to be significant (P < 0.1)
(Tab. V). In the candidate models with CP
plus NSC2 or CP2 plus NSC2 the parameter,
method of conservation
(P < 0.1)
method of conservation
(P < 0.1)
group of forage species
(P < 0.1)
group of forage species
(P < 0.1)
NSC * references
NSC * animal species
NSC2 * group of forage
NSC2 * group of forage
NDF * references
CP2 + NSC + references +
animal species + interactions
CP + NSC2 + interactions
with group of forage species
CP2 + NSC2 + interactions
with group of forage species
CP2 + NDF + references
NDF2 * animal species
NDF2 * methodology
CP + NDF2 + animal species
+ methodology + interactions
group of forage species, tended to be
significant (P < 0.1), although the interactions
between the group of forage species and
these chemical components were
significant (P < 0.05) (Tab. V).
In all candidate models CP or CP2 were
significant after stepwise regression
(Tab. V). Most candidate models could not
be used, because the parameters, references
or methodology were significant after
stepwise regression. These parameters were not
significant in the models with CP, CP2, CP2
plus MC, CP plus NSC2 and with CP2 plus
NSC2. Neither the prediction with CP2 nor
the prediction with CP2 plus MC or NSC2
were better than the prediction with only
CP (Tab. VI). In these models, MSPE were
for 100% due to the disturbance and the
probability of the estimates, MC or NSC2,
tended to be significant (P < 0.1).
Nevertheless a model with CP2 plus MC
or NSC2 tended to predict YMP more
precisely than a model with only CP, because
these models had a higher percentage of
deviations (predicted minus observed values)
within the envelope of acceptable precision
with limits of 1.5*SD (Tab. VII, Figs. 1a
CP2 and MC were almost orthogonal,
because the sum of the squares (regression)
of the model with CP2 plus MC was 313,
with only CP2 was 275 and with only MC
was 68, as well as regression coefficients of
CP2 were similar between the model with
CP2 plus MC and the model with CP2. The
parameter group of forage species did not
improve the model with CP2 plus NSC2
because of multicollinearity and interactions
with CP2 or NSC2.
4.1. Duodenal flow of microbial protein and chemical components
CP was the most important chemical
component in the estimation of YMP. CP
expresses the availability of N for the
microbes in the rumen and is positively related
to YMP and EMPS as long as nitrogen is
not limiting and the protein is not used as a
source of energy [
]. NSC had an extra
effect on YMP, because of the energy
supply. An increasing amount of available
NSC in the rumen can prevent the use of CP
as a source of energy for microbial growth.
P estimate 5.33 + 0.0393 * CP
P estimate < 0.05 < 0.05
P estimate 8.06 + 0.000125 * CP2
P estimate < 0.05 < 0.05
a MC = method of conservation; for abbreviations, see Table II.
% of deviations inside the envelope of acceptable precision
However, NSC can have a negative
influence on the rumen function [
limiting effect of NSC on YMP was found in
this study, which was a consequence of the
use of rations with only forages.
NSC could be replaced by NDF in the
prediction of YMP. NDF is important for
the rumen function and environment,
because NDF does not only have a
mechanical function, stimulating rumination and
forming a mat in the rumen, but also a
biochemical function because of the
stimulation of salivation and the buffering capacity
]. NDF had a decreasing effect on YMP,
because a low concentration of NDF in dry
matter coincides with a high digestibility of
forages and high concentrations of NSC
and CP in dry matter. Parallel to this, a low
concentration of NDF in DM means a high
digestion rate of NDF [
], which affects
the rate of digestion of carbohydrates [
NDF content is also an indicator for the
maturity of forages and for the difference
between legumes and grasses [
4.2. Duodenal flow of microbial protein and other parameters
When MC was included in the model
with CP2, the prediction of YMP was more
precise. MC has different effects on the
microbial protein synthesis in the rumen. The
duodenal flow of microbial protein was
higher for fresh forages than for other
methods of conservation, which agreed with the
observations of Holden et al. [
] in an
experiment with dairy cows fed Orchard
grass. The lower values for silage is a
consequence of its lower proportions of
water-soluble carbohydrates [
carbohydrates are energy, which is rapidly
available for the microbial growth in the
rumen. The lower values for hay and dried
forages may be the result of a decreased rate
of ruminal degradation of dietary CP, which
diminished the availability of N for
microbes in the rumen [
A group of forage species tended to have
an effect on YMP, but had interactions with
CP2 and NSC2. The reason for these
interactions is that the content of these chemical
components as well as YMP differed
significantly between legumes and grasses
(Tab. II). Another reason can be a different
slope in the effect of CP content or NSC
content on YMP between legumes and
grasses, because legumes have a lower
digestibility of the cell walls than grasses
]. This difference was not significant in
this study because of the small numbers in
the data file.
In some models, animal species were
significant in the prediction of YMP (Tab. V).
These models were not useful, because
references or methodology were also
significant. A difference in YMP between cattle
and sheep was expected, because they
differ in rumen digestion and passage rates
It is noteworthy that the other
parameters, which were not significant in the
prediction of YMP, may also influence the
rates of degradation and passage in the
rumen. These parameters, such as the
physical form of presentation and DMI, are
known to influence microbial protein
synthesis. Chopping has a positive effect on
DMI through a decreased fill effect and an
increased passage rate [
10, 15, 37
efficiency of microbial protein synthesis is
positively related to the rumen passage rate
as a result of the reducing internal turnover
of microbes and reducing maintenance cost
for bacterial growth [
]. The effect of
DMI on the passage rate may partly be
represented by NDF in the prediction
equations, since NDF content is well related to
DMI and gastrointestinal fill .
However, the influence of chopping and DMI
would have been greater, if the data file did
contain diets with ground forages and no
restricted DMI (90% of ad lib).
The parameter methodology was
significant in some models. The main differences
between in vivo trials originate from the
variation in the methods used for measuring
duodenal flow and partitioning protein in
microbial versus dietary origin [
18, 19, 51
The parameter references were also
significant in some models, due to the
heterogeneous origin of the data.
The statistical parameters were poor, the
percentages of deviations of the predictions
within the envelope of acceptable precision
were lower than 95%, R2 was low and the
prediction error or coefficient of variation
(CV) was high. CV was about 30% and
close to the CV (26.3%) of the best fit
equation of Oldick et al. [
]. This equation
estimates the daily duodenal flow of microbial
N from DMI and NDF and is composed on
the basis of a data file containing 213
treatments with cattle fed mixed rations.
4.3. Validation of the PDI-system
The statistics of the validation of the
calculation from the PDI-system [
compared with the statistics of the
regressions from this study on the data file of the
present study. The PDI calculation was
composed using a data file with sheep and
cattle and mixed diets and the duodenal
flow of microbial N (g·d–1) was calculated
as FOM*23.2 microbial N (g·kg–1 FOM).
FOM is fermentable OM calculated from
OM digested in the total tract (DOM) minus
bypass protein, volatile fatty acids and
alcohol in silage, and lipids. The values of the
PDI calculation were divided with the daily
DM intake (kg·d–1), to obtain the duodenal
flow of microbial N per kg of DM intake.
This calculation excludes the great effect of
the daily intake of DM or OM on the daily
flow of microbial N (Tab. IV).
When the values of the PDI calculation
were related to the YMP values of the data
file, R2 was very low (0.10), the prediction
error was 36% and MSPE was 92% due to
disturbance. The percentage of deviations
inside the envelope of acceptable precision
] was also lower than 95% (Tab. VII,
Fig. 1c). Generally the statistics of the
validation of the PDI calculation were similar
to the statistics of the regressions from this
The chemical composition of forages,
with or without the method of conservation,
is a poor indication for the duodenal flow of
microbial N per kg DM intake (YMP) in
ruminants fed diets with forages only. The
precision of the validation of the PDI
prediction was close to the precision of the
regressions of YMP from this study. The
equations from this study need validations
with other independent data sets.
Predicting YMP, the yield of microbial
protein, is more difficult than the prediction
of the daily duodenal flow of microbial
protein from DM intake. The prediction of
YMP partly implies EMPS, which depends
on quantitative, qualitative and dynamic
factors of animal and dietary origin. These
factors are necessary to improve the
predictions of this study and their precision. To
integrate all these factors to predict the
duodenal flow of microbial N per day or per
kg of DM intake, mechanistic rumen
models are proposed [
The authors gratefully thank Dr. R. Vérité
(INRA, centre de Rennes, Unité Mixte de
Recherches Production du Lait, 35590
SaintGilles, France) for his scientific contributions.
 Bauchart D. , Doreau M. , Legay-Carmier F. , Utilisation digestive des lipides et conséquences de leur introduction sur la digestion du ruminant , Bull. Tech. CRZV Theix 61 ( 1985 ) 65 - 77 .
 Beever D.E. , Thomson D.J. , Cammell S.B. , Harrison D.G. , The digestion by sheep of silages made with and without the addition of formaldehyde , J. Agric. Sci . (Camb.) 88 ( 1977 ) 61 - 70 .
 Berzaghi P. , Herbein J.H. , Polan C.E. , Intake, site, and extent of nutrient digestion of lactating cows grazing pasture , J. Dairy Sci . 79 ( 1996 ) 1581 - 1589 .
 Bibby J. , Toutenberg H. , Prediction and improvement estimation in linear models , John Willey & Sons, London, UK, 1977 .
 Brake A.C. , Goetsch A.L. , Forster L.A. Jr. , Landis K.M. , Feed intake, digestion and digesta characteristics of cattle fed bermudagrass or orchardgrass alone or with ground barley or corn , J. Anim. Sci . 67 ( 1989 ) 3425 - 3436 .
 Caton J.S. , Erickson D.O. , Carey D.A. , Ulmer D.L. , Influence of Aspergillus oryzae fermentation extract on forage intake, site of digestion, in situ degradability, and duodenal amino acid flow in steers grazing cool-season pasture , J. Anim. Sci . 71 ( 1993 ) 779 - 787 .
 Charmley E. , Veira D.M. , Inhibition of proteolysis at harvest using heat in alfalfa silages: effects on silage composition and digestion by sheep , J. Anim. Sci . 68 ( 1990 ) 758 - 766 .
 Charmley E. , Veira D.M. , Inhibition of proteolysis in alfalfa silages using heat at harvest: effects on digestion in the rumen, voluntary intake and animal performance , J. Anim. Sci . 68 ( 1990 ) 2042 - 2051 .
 Clark J.H. , Klusmeyer T.H. , Cameron M.R. , Microbial protein synthesis and flows of nitrogen fractions to the duodenum of dairy cows , J. Dairy Sci . 75 ( 1992 ) 2304 - 2323 .
 Colucci P.E. , Macleod G.K. , Grovum W.L. , McMillan I. , Barney D.J. , Digesta kinetics in sheep and cattle fed diets with different forage to concentrate ratios at high and low intakes , J. Dairy Sci . 73 ( 1990 ) 2143 - 2156 .
 Colucci P.E. , Macleod G.K. , Grovum W.L. , Cahill L.W. , McMillan I. , Comparative digestion in sheep and cattle fed different forage to concentrate ratios at high and low intakes , J. Dairy Sci . 72 ( 1989 ) 1774 - 1785 .
 Demarquilly C. , Fertilisation et qualité du fourrage, Fourrages 69 ( 1977 ) 61 - 84 .
 Demeyer D. , Van Nevel C. , Influence of substrate and microbial interaction on efficiency of rumen microbial growth , Reprod. Nutr . Dév. 26 ( 1986 ) 161 - 179 .
 Dijkstra J. , Neal H.D. , Beever D.E. , France J., Simulation of nutrient digestion, absorption and outflow in the rumen: model description , J. Nutr . 122 ( 1992 ) 2239 - 2256 .
 Djouvinov D.S. , Todorov N.A. , Influence of dry matter intake and passage rate on microbial protein synthesis in the rumen of sheep and its estimation by cannulation and a non-invasive method , Anim. Feed Sci. Technol . 48 ( 1994 ) 289 - 304 .
 Elizalde J.C. , Merchen N.R. , Faulkner D.B. , Supplemental cracked corn for steers fed fresh alfalfa. I. Effects on digestion of organic matter, fiber, and starch , J. Anim. Sci . 77 ( 1999 ) 457 - 466 .
 Elizalde J.C. , Merchen N.R. , Faulkner D.B. , Supplemental cracked corn for steers fed fresh alfalfa. II. Protein and amino acid digestion , J. Anim. Sci . 77 ( 1999 ) 467 - 475 .
 Faichney G.J. , Digesta flow , in: Forbes J.M., France J . (Eds.), Quantitative aspects of ruminant digestion and metabolism , CAB International, Wallingford, UK, 1993 , pp. 53 - 85 .
 Firkins J.L. , Allen M.S. , Oldick B.S. , St-Pierre N.R. , Modeling ruminal digestibility of carbohydrates and microbial protein flow to the duodenum , J. Dairy Sci . 81 ( 1998 ) 3350 - 3369 .
 Galloway D.L. Sr. , Goetsch A.L. , Forster L.A. Jr. , Brake A.C. , Johnson Z.B., Digestion, feed intake, and live weight gain by cattle consuming bermudagrass and supplemented with different grains , J. Anim. Sci . 71 ( 1993 ) 1288 - 1297 .
 Galloway D.L. Sr. , Goetsch A.L. , Sun W. , Forster L.A. Jr. , Murphy G.E. , Grant E.W. , Johnson Z.B., Digestion, feed intake, and live weight gain by cattle consuming bermudagrass hay supplemented with whey , J. Anim. Sci . 70 ( 1992 ) 2533 - 2541 .
 Genstat , Reference manual, Release 5 .2. VSN International, Oxford, UK, 2000 .
 Goering H.K. , Van Soest P.J. , Forage fiber analyses, Agricultural Handbook No. 379 , US Department of Agriculture, Washington, DC, 1970 , pp. 1 - 20 .
 Goetsch A.L. , Forster L.A. Jr. , Murphy G.E. , Grant E.W. , Galloway D.L. Sr. , West C.P. , Digestion and live-weight gain by beef cattle consuming bermudagrass supplemented with grain or different high-protein foodstuffs , Anim. Prod . 51 ( 1990 ) 263 - 275 .
 Hogan J.P. , Lindsay J.R. , The digestion of nitrogen associated with plant cell wall in the stomach and small intestine of the sheep , Aust. J. Agric. Res . 31 ( 1979 ) 147 - 153 .
 Holden L.A. , Muller L.D. , Varga G.A. , Hillard P.J. , Ruminal digestion and duodenal nutrient flows in dairy cows consuming grass as pasture, hay, or silage , J. Dairy Sci . 77 ( 1994 ) 3034 - 3042 .
 Hoover W.H. , Stokes S.R. , Balancing carbohydrates and proteins for optimum rumen microbial yield , J. Dairy Sci . 74 ( 1991 ) 3630 - 3644 .
 Hume I.D. , Purser D.B. , Ruminal and postruminal protein digestion in sheep fed on subterranean clover harvested at four stages of maturity, Aust . J. Agric. Res . 26 ( 1975 ) 199 - 208 .
 Jones A.L. , Goetsch A.L. , Effects of dietary forage proportion on digestive function in maintenance-fed beef cows. 1. Fescue and clover hays , Arch. Anim. Nutr . 37 ( 1987 ) 685 - 699 .
 Jones A.L. , Goetsch A.L. , Stokes S.R., Effects of dietary forage proportion on digestive function in maintenance-fed beef cows. 2. Fescue and bermudagrass hays , Arch. Anim. Nutr . 37 ( 1987 ) 701 - 711 .
 Jones A.L. , Goetsch A.L. , Stokes S.R. , Coldberg M., Effects of dietary forage proportion on digestive function in maintenance-fed beef cows. 3. Bermudagrass and clover hays , Arch. Anim. Nutr . 37 ( 1987 ) 1009 - 1020 .
 Kawas J.R. , Jorgensen N.R. , Lu C.D. , Influence of alfalfa maturity on feed intake and site of nutrient digestion in sheep , J. Anim. Sci . 68 ( 1990 ) 4376 - 4386 .
 Krysl L.J. , Branine M.E. , Cheema A.U. , Funk M.A. , Galyean M.L. , Influence of soybean meal and sorghum grain supplementation on intake, digesta kinetics, ruminal fermentation, site and extent of digestion and microbial protein synthesis in beef steers grazing blue grama rangeland , J. Anim. Sci . 67 ( 1989 ) 3040 - 3051 .
 Krysl L.J. , Judkins M.B. , Bohman V.R. , Influence of ruminal or duodenal soybean oil infusion on intake, ruminal fermentation, site and extent of digestion, and microbial protein synthesis in beef heifers consuming grass hay , J. Anim. Sci . 69 ( 1991 ) 2585 - 2590 .
 Lindsay J.R. , Hogan J.P. , Digestion of two legumes and rumen bacterial growth in defaunated sheep , Aust. J. Agric. Res . 23 ( 1972 ) 321 - 330 .
 Makoni N.F. , von Keyserlingk M.A.G. , Shelford J.A. , Fisher L.J. , Degradability of frozen and ensiled alfalfa proteins by sheep and assessment of duodenal digesta protein , Anim. Feed Sci. Technol . 53 ( 1995 ) 221 - 231 .
 Malbert C.H. , Baumont R. , The effects of intake of lucerne (Medicago sativa L.) and orchard grass (Dactylis glomerata L.) hay on the motility of the forestomach and digesta flow at the abomaso-duodenal junction of the sheep , Brit. J. Nutr . 61 ( 1989 ) 699 - 714 .
 Merchen N.R. , Bourquin L.D. , Process of digestion and factors influencing digestion of forage-based diets by ruminants , in: Fahey G.C. Jr. (Ed.), Forage quality, evaluation, and utilization , ASA, CSSA, SSSA , Madison, WI , USA, 1994 , pp. 564 - 612 .
 Merchen N.R. , Satter L.D. , Digestion of nitrogen by lambs fed alfalfa conserved as baled hay or as low moisture silage , J. Anim. Sci . 56 ( 1983 ) 943 - 951 .
 Mitchell P.L. , Misuse of regression for empirical validation of models, Agric . Syst. 54 ( 1997 ) 313 - 326 .
 Muntifering R.B. , Wedekind K.J. , Knifley T. , Ely D.G. , Effects of processing on the supplemental protein value of distillers byproducts in forage diets , J. Anim. Sci . 61 ( 1985 ) 647 - 653 .
 Narasimhalu P. , Teller E. , Vanbelle M. , Foulon M. , Dasnoy F. , Apparent digestibility of nitrogen in rumen and whole tract of friesian cattle fed direct-cut and wilted grass silages , J. Dairy Sci . 72 ( 1989 ) 2055 - 2061 .
 O 'Mara F.P. , Stakelum G.K. , Dillon P. , Murphy J.J. , Rath M. , Rumen fermentation and nutrient flows for cows fed grass and grass supplemented with molassed beet pulp pellets , J. Dairy Sci . 80 ( 1997 ) 2466 - 2474 .
 Oldick B.S. , Firkins J.L. , St-Pierre N.R. , Estimation of microbial nitrogen flow to the duodenum of cattle based on dry matter intake and diet composition , J. Dairy Sci . 82 ( 1999 ) 1497 - 1511 .
 Perez J.F. , Balcells J. , Guada J.A. , Castrillo C. , Determination of rumen microbial-nitrogen production in sheep: a comparison of urinary purine excretion with methods using 15N and purine bases as markers of microbial-nitrogen entering the duodenum , Brit. J. Nutr . 75 ( 1996 ) 699 - 709 .
 Peyraud J.L. , Astigarraga L. , Faverdin P. , Digestion of fresh perennial ryegrass fertilized at two levels of nitrogen by lactating dairy cows , Anim. Feed Sci. Technol . 64 ( 1997 ) 155 - 171 .
 Poncet C. , Michalet-Doreau B. , McAllister T. , Rémond D. , Dietary compounds escaping rumen digestion , in: Journet M., Grenet E. , Farce M.-H. , Thériez M . (Eds.), Recent developments in the nutrition of herbivores, INRA éditions , Paris, France, 1995 .
 Robertson J.B., Van Soest P.J., The detergent system of analysis and its application to human foods , in: James W.P.T., Theander O . (Eds.), The analysis of dietary fiber in food , Marcel Dekker, New York, USA, 1981 , pp. 123 - 158 .
 Sauvant D. , Grenet E. , Doreau M. , Dégradation chimique des aliments dans le réticulo-rumen: cinétique et importance , in: Jarrige R., Ruckebusch Y. , Demarquilly C. , Farce M.-H. , Journet M . (Eds.), Nutrition des ruminants domestiques - Ingestion et digestion , INRA éditions, Paris, France, 1995 , pp. 383 - 406 .
 Sauvant D. , Dijkstra J. , Mertens D. , Optimisation of ruminal digestion: a modelling approach , in: Journet M., Grenet E. , Farce M.-H. , Thériez M . (Eds.), Recent developments in the nutrition of herbivores, INRA éditions , Paris, France, 1995 .
 Stern M.D. , Varga G.A. , Clark J.H. , Firkins J.L. , Huber J.T. , Palmquist D.L. , Evaluation of chemical and physical properties of feeds that affect protein metabolism in the rumen , J. Dairy Sci . 77 ( 1994 ) 2762 - 2786 .
 Stokes S.R. , Goetsch A.L. , Jones A.L. , Landis K.M. , Feed intake and digestion by beef cows fed prairie hay with different levels of soybean meal and receiving postruminal administration of antibiotics , J. Anim. Sci . 66 ( 1988 ) 1778 - 1789 .
 Stokes S.R. , Goetsch A.L. , Landis K.L. , Feed intake and digestion by beef steers consuming and receiving ruminal insertions of prairie hay differing in level and particle size , J. Anim. Sci . 66 ( 1988 ) 1267 - 1274 .
 Tamminga S. , Van Straalen W.M. , Subnel A.P.J. , Meijer R.G.M. , Steg A. , Wever C.J.G. , Blok M.C. , The Dutch protein evaluation system: the DVE/OEB-system , Livest. Prod. Sci . 40 ( 1994 ) 139 - 155 .
 Thompson D.J. , Beever D.E. , Lonsdale C.R. , Haines M.J. , Cammell S.B. , Austin A.R. , The digestion by cattle of grass silage made with formic acid and formic acid-formaldehyde, Brit . J. Nutr . 46 ( 1981 ) 193 - 207 .
 Tjandraatmadja M. , MacRae I.C. , Norton B.W. , Digestion by sheep of silages prepared from mixtures of tropical grasses and legumes , J. Agric. Sci . (Camb.) 120 ( 1993 ) 407 - 415 .
 Van Soest P.J. , Nutritional ecology of the ruminant , Cornell University Press, New York, USA, 1982 .
 Van Soest P.J. , Robertson J.B. , Lewis B.A. , Methods for dietary fiber, neutral detergent fiber, and nonstarch polysaccharides in relation to animal nutrition , J. Dairy Sci . 74 ( 1991 ) 3583 - 3597 .
 Van Soest P.J. , Wine R.H. , Use of detergents in the analysis of fibrous feeds. IV. Determination of plant cell wall constituents , J. Assoc. Off. Anal. Chem . 50 ( 1967 ) 50 - 55 .
 Van Vuuren A.M. , Krol-Kramer F ., van der Lee R.A. , Corbijn H. , Protein digestion and intestinal amino acids in dairy cows fed fresh Lolium perenne with different nitrogen contents , J. Dairy Sci . 75 ( 1992 ) 2215 - 2225 .
 Varel V.H. , Kreikemeier K.K. , Influence of feeding Aspergillus oryzae fermentation extract (Amaferm) on in situ fiber degradation, ruminal fermentation, and microbial protein synthesis in nonlactating cows fed alfalfa or bromegrass hay , J. Anim. Sci . 72 ( 1994 ) 1814 - 1822 .
 Vérité R. , Michalet-Doreau B. , Chapoutot P. , Peyraud J.-L. , Poncet C. , Révision du système des protéines digestibles dans l'intestin ( P.D.I .), Bull. Tech. CRZV Theix 70 ( 1987 ) 19 - 34 .
 Walker D.J. , Egan A.R. , Nader C.J. , Ulyatt M.J. , Storer G.B. , Rumen microbial protein synthesis and proportions of microbial and non-microbial nitrogen flowing to the intestines of sheep, Aust . J. Agric. Res . 26 ( 1975 ) 609 - 708 .